The paper discusses the critical elements of brain-computer interface (BCI) systems, focusing on BCI paradigms and neural coding. It highlights the importance of these elements in BCI research and reviews the existing main BCI paradigms and neural coding models. The authors define BCI paradigms as specific mental tasks or external stimuli designed to represent user intentions, and neural coding as the process of encoding these intentions into brain signals. They outline the principles for designing BCI paradigms, emphasizing user-centered design and the importance of good separability, ease of performance, safety, and user experience. The paper also reviews various BCI paradigms, including motor imagery (MI), steady-state visual evoked potential (SSVEP), and P300 paradigms, and discusses the neural coding mechanisms under different paradigms. It explores the relationship between BCI paradigm, neural coding, and neural decoding, and the role of brain imaging techniques such as intracortical local field potentials (LFP), electroencephalogram (EEG), functional near-infrared spectroscopy (fNIRS), functional magnetic resonance imaging (fMRI), magnetoencephalography (MEG), and hybrid BCI (hBCI). The paper concludes by discussing challenges and future research directions, including user-centered design, revolutionizing traditional BCI paradigms, improving brain signal collection techniques, and integrating BCI with advanced AI technology to enhance decoding performance.The paper discusses the critical elements of brain-computer interface (BCI) systems, focusing on BCI paradigms and neural coding. It highlights the importance of these elements in BCI research and reviews the existing main BCI paradigms and neural coding models. The authors define BCI paradigms as specific mental tasks or external stimuli designed to represent user intentions, and neural coding as the process of encoding these intentions into brain signals. They outline the principles for designing BCI paradigms, emphasizing user-centered design and the importance of good separability, ease of performance, safety, and user experience. The paper also reviews various BCI paradigms, including motor imagery (MI), steady-state visual evoked potential (SSVEP), and P300 paradigms, and discusses the neural coding mechanisms under different paradigms. It explores the relationship between BCI paradigm, neural coding, and neural decoding, and the role of brain imaging techniques such as intracortical local field potentials (LFP), electroencephalogram (EEG), functional near-infrared spectroscopy (fNIRS), functional magnetic resonance imaging (fMRI), magnetoencephalography (MEG), and hybrid BCI (hBCI). The paper concludes by discussing challenges and future research directions, including user-centered design, revolutionizing traditional BCI paradigms, improving brain signal collection techniques, and integrating BCI with advanced AI technology to enhance decoding performance.